Intakes of 4 dietary lignans and cause-specific and all-cause mortality in the Zutphen Elderly Study

¹ From the Centre for Nutrition and Health, National Institute for Public Health and the Environment, Bilthoven, Netherlands (IEJM, EJMF, and HBB); RIKILT–Institute of Food Safety, Wageningen University and Research Centre, Wageningen, Netherlands (ICWA and PCHH); and the Division of Human Nutrition, Wageningen University, Wageningen, Netherlands (EJMF and DK)

² Supported by grant no. 014-12-014 (Nutrition: Health, Safety, and Sustainability) from the Netherlands Organisation for Health Research and Development.

³ Reprints not available. Address correspondence to IEJ Milder, Centre for Nutrition and Health, National Institute for Public Health and the Environment, PO Box 1, 3720 BA Bilthoven, Netherlands. E-mail: ivon.milder{at}rivm.nl.

ABSTRACT  
Background: Plant lignans are converted to enterolignans that have antioxidant and weak estrogen-like activities, and therefore they may lower cardiovascular disease and cancer risks.

Objective: We investigated whether the intakes of 4 plant lignans (lariciresinol, pinoresinol, secoisolariciresinol, and matairesinol) were inversely associated with coronary heart disease (CHD), cardiovascular diseases (CVD), cancer, and all-cause mortality.

Design: The Zutphen Elderly Study is a prospective cohort study in which 570 men aged 64–84 y were followed for 15 y. We recently developed a database and used it to estimate the dietary intakes of 4 plant lignans. Lignan intake was related to mortality with the use of Cox proportional hazards analysis.

Results: The median total lignan intake in 1985 was 977 µg/d. Tea, vegetables, bread, coffee, fruit, and wine were the major sources of lignan. The total lignan intake was not related to mortality. However, the intake of matairesinol was inversely associated with CHD, CVD, and all-cause mortality (P 0.05 for all) and cancer (P = 0.06). Multivariate-adjusted rate ratios (95% CI) per 1-SD increase in intake were 0.72 (0.53, 0.98) for CHD, 0.83 (0.69, 1.00) for CVD, 0.86 (0.76, 0.97) for all-cause mortality, and 0.81 (0.65, 1.00) for cancer.

Conclusions: Total lignan intake was not associated with mortality. The intake of matairesinol was inversely associated with mortality due to CHD, CVD, cancer, and all causes. We cannot exclude the possibility that the inverse association between matairesinol intake and mortality is due to an associated factor, such as wine consumption.

Key Words: Lignans • diet • phytoestrogens • secoisolariciresinol • matairesinol • lariciresinol • pinoresinol • cancer • cardiovascular disease • mortality • cohort study • elderly • Netherlands

INTRODUCTION  
Lignans are biphenolic compounds in plant foods that belong to the phytoestrogen group. Tea, coffee, whole-grain products, vegetables, fruit, nuts and seeds are dietary sources of lignans (1, 2). Flaxseed has an exceptionally high lignan concentration. Plant lignans can be converted by intestinal bacteria into the so-called enterolignans enterolactone and enterodiol. It has been shown that, besides the known enterolignan precursors secoisolariciresinol (SECO) and matairesinol (MAT), lariciresinol (LARI) and pinoresinol (PINO) were efficiently converted (3).

There are several mechanisms by which phytoestrogens may protect against cardiovascular disease (CVD) and cancer. Enterolignans have weak estrogen-like activity. They may bind to estrogen receptors and ß (4), which are expressed in various tissues. They also exert an effect via estrogen receptor–nonmediated mechanisms (5), such as the inhibition of several enzymes (eg, aromatase and 5-reductase) and the stimulation of the production of sex hormone–binding globulin (SHBG). In addition, plant lignans and, to a lesser extent, enterolignans have antioxidant activity (6).

Epidemiologic evidence of the potential protective effects of lignan intake is limited. Previous studies included only SECO and MAT in relation to CVD, and these studies yielded conflicting results (7, 8, 9, 10). The relation between lignan intake and cancer has mainly been studied with respect to hormone-dependent cancers. In women, protection against ovarian, endometrial, and thyroid cancer was observed in case-control studies (11). Conflicting results were obtained for the relation between intake of SECO plus MAT and breast cancer (11, 12, 13). In men, intake of SECO plus MAT was not associated with testicular (14) or prostate (15) cancer in case-control studies. Intake of SECO plus MAT was significantly inversely associated with lung cancer in men but not in women (16).

Previously, data on only SECO and MAT in foods were available. We have developed a liquid chromatography–mass spectrometry method to measure LARI and PINO in addition to SECO and MAT in foods and beverages (17), and we constructed a lignan database of plant foods commonly consumed in the Netherlands (2). With this database, we evaluated whether a higher intake of plant lignans is associated with a lower risk of CHD, CVD, cancer, and all-cause mortality in a cohort of Dutch elderly men.

SUBJECTS AND METHODS  
Study population
The Zutphen Elderly Study is a prospective cohort study on risk factors for chronic diseases in elderly men. It is an extension of the Zutphen Study, the Dutch contribution to the Seven Countries study (18). In 1985, men from the original Zutphen cohort who were still alive (n = 554) were invited to participate in the Zutphen Elderly Study. In addition, a random sample of men of the same age who were living in Zutphen but who had not been included in the original cohort was selected. In total, 1266 men aged 65–84 y were invited, of which 939 (74%) agreed to participate (19). Complete information on baseline diet, prevalence of chronic disease, and other important risk factors was available for 787 men. We excluded men with a history of myocardial infarction (n = 97), stroke (n = 25), heart failure (n = 28), diabetes (n = 41) or cancer (n = 56). One man consumed flaxseed as a supplement and was excluded because of his extremely high lignan intake (28 mg/d in 1985). After these exclusions 570 men remained in the analyses. The study was approved by the Medical Ethics Committee of the University of Leiden, The Netherlands. Informed consent was obtained from all participants.

Data collection
Dietary surveys and medical examinations were conducted between March and June in 1985, 1990, and 1995. Information on the habitual diet of the participants was collected by experienced dietitians who used a cross-check dietary history method adapted to the usual Dutch diets (20). Participants were interviewed, in the presence of the person who usually prepared the meals, about their usual food consumption on weekdays and weekends. The estimated consumption of foods during 1 d or 1 wk was compared with the estimated quantities of food bought for the whole family. Discrepancies were discussed with the participant and resolved accordingly. The food consumption data were encoded by the dietitians and converted into energy and nutrients with the use of the Dutch Food Composition Table from 1985 that was updated with 2001 data for ß-carotene, folate, vitamin E and dietary fiber. The intakes of trans fatty acids, -linolenic acid, eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) were estimated previously by Oomen et al (21); the intake of flavonols and flavones were estimated by Hertog et al (19); and the intakes of catechins were estimated by Arts et al (22). Lignan intake was estimated by using the abovementioned database (2) that included the LARI, PINO, SECO, and MAT contents of commonly consumed Dutch plant foods. Lignans were measured after the hydrolysis of lignan glycosides, and thus the lignan contents included both aglycones and glycosides. In this report, the lignan contents are expressed as the aglycone weight of the molecule.

Trained physicians measured height and weight, and blood pressure measurements were taken while the subjects were lying in a supine position. Nonfasting serum total and HDL cholesterol measurements were taken enzymatically (23). Physical activity was estimated using a validated questionnaire, originally designed for retired men (24). It included questions on walking, bicycling, hobbies, gardening, odd jobs and sports. Information on smoking was obtained by using a standard questionnaire and on alcohol consumption by the cross-check dietary history method.

Case ascertainment
Municipal registries were checked at 5-y intervals from 1985 to July 2000 to ascertain whether participants had died in the interim. Causes of death were obtained from various sources, mainly hospital discharge data, and supplemental information was obtained from the Netherlands Cancer Registry, general practitioners, and, until 1990, from Statistics Netherlands (Voorburg, Netherlands). All information was used to establish the contributory causes of death. Because the cause of death is often difficult to ascertain in elderly people, the secondary and tertiary causes of death, as well as the primary cause, were used in the statistical analyses. The final coding of the causes of death was performed by one clinical epidemiologist, according to the 9th revision of the International Classification of Causes of Death (ICD-9). We classified ICD codes 410–414 as CHD, codes 390–459 as CVD, and codes 140–208 as cancer. One man was lost to follow-up in 1991, and his follow-up was censored at 30 June 1991. Information on the causes of death was missing for 2 men, and their follow-up was censored at their date of death.

Statistical analysis
The subjects were divided into tertiles of total lignan intake at baseline. To test for differences in baseline characteristics, we used analysis of variance for normally distributed variables, the Kruskal-Wallis test for skewed distributed variables, and the chi-square test for categorical variables. Spearman's rank order correlations between the intake of lignans and the intakes of other nutrients and lignan-containing foods were calculated. Rate ratios (RRs) and 95% CIs per 1-SD increase in intakes of total and individual lignans were estimated by hazard ratios obtained from Cox proportional hazards models. For comparison with the literature, we also determined the RRs per 1-SD increase in intakes of SECO plus MAT.

To represent long-term exposures, we ascertained the cumulative averages of lignan intake, other dietary variables, and physical activity (25). Thus, the mortality from 1985 to 1990 was related to the estimated values in 1985, and that between 1990 and 1995 and between 1995 and 2000 was related to the average of the values in 1985 and 1990 and the average of values in 1985, 1990 and 1995, respectively. The follow-up rates for the dietary questionnaires in 1990 and 1995 were 82% and 65%, respectively, of the men who were still alive at those dates. When updated dietary data were not available, the last available value was used. For smoking, the most recent data were used. Smoking status was included in the models as 3 categories (current, former, or never); duration and amount were calculated from the questionnaire and entered as a continuous variable.

Besides age-adjusted analyses, we used 2 multivariate models. In model 1, we adjusted for age, smoking status, amount and duration of smoking, physical activity, and energy intake. In model 2 (full model), we also adjusted for intakes of dietary fiber and ß-carotene.

Initially, BMI, alcohol consumption (>30g/d, 30 g/d, or no alcohol), socioeconomic status (4 categories based on occupation), and intakes of saturated fatty acids, trans fatty acids, -linolenic acid, EPA, DHA, dietary cholesterol, vitamins C and E, and folate were included as potential confounders, but they were removed because they had P values > 0.3 for each endpoint. Re-entering the previously removed variables one by one in model 2 did not change any of the RRs (<5% change).

Smokers have greater oxidative stress than do nonsmokers, and thus the antioxidant effects of lignans may be more marked in the former than in the latter. We therefore performed analyses stratified for smoking status (current, former, or never). The intake of dietary fiber may influence the bioavailability of lignans, perhaps because of effects on intestinal transit time. Therefore, we also stratified for fiber intake above or below the median (23 g/d). All statistical analyses were conducted using SAS statistical analysis package (version 9.1; SAS Institute Inc, Cary, NC).

RESULTS  
The median total intake of lignans in 1985 was 977 µg/d (mean 1037 µg/d) (Table 1). Of the individual lignans, LARI contributed the most (48%) to mean lignan intake, and PINO (36%), SECO (15%), and MAT (1%) followed. The intakes of the individual lignans were moderately to highly correlated (Spearman r = 0.4–0.8 P < 0.001). The major sources of lignan for this population were tea (28%), vegetables (27%), bread (14%), coffee (10%), fruit (9%), and wine (1%). The median lignan intake was 1000 µg/d in 1990 and 1053 µg/d in 1995. Spearman's correlation coefficient between total lignan intake in 1985 and 1990 was 0.51 and that between total lignan intake in 1985 and 1995 was 0.47 (P < 0.001 for both).

View this table:
TABLE 1. Contribution of different food groups to lignan intake in 1985 in 570 men in the Zutphen Elderly Study¹

 
Men in the highest tertile of total lignan intake were less likely to smoke, were more physically active, and had a higher energy intake than were men in the lowest tertile (Table 2). After adjustment for energy intake, total lignan intake was positively correlated with intakes of fiber (Spearman r = 0.51), vitamin C (Spearman r = 0.39), folate (Spearman r = 0.37), and ß-carotene (Spearman r = 0.32) and was weakly and inversely associated with the intakes of saturated fatty acids (Spearman r = –0.13), trans fatty acids (Spearman r = –0.09), and dietary cholesterol (Spearman r = –0.10). Tea was the major source of lignan. As a consequence, lignan intake was correlated with intakes of flavonols and flavones (Spearman r = 0.70) and catechins (Spearman r = 0.62), because in this population tea also is the major source of these bioactive compounds.

View this table:
TABLE 2. Characteristics of 570 men in the Zutphen Elderly Study by tertile (T) of total lignan intake in 1985¹

 
After 15 y of follow-up (5888 person-years), 392 of the 570 men had died. Of these 392 men, 84, 197, and 125 had CHD, CVD, and cancer, respectively, as the primary, secondary, or tertiary cause of death. After multivariate adjustment total lignan intake was not associated with cause-specific or all-cause mortality (Table 3). To compare our results with previous data, we have calculated the RRs per SD (55 µg/d) increase in the intake of SECO plus MAT. The RRs (95% CI) after full adjustment were 0.81 (0.58, 1.12) for CHD, 0.86 (0.69, 1.06) for CVD, 0.87 (0.68, 1.11) for cancer, and 0.95 (0.83, 1.09) for all-cause mortality. We did not find significant lignan intake x smoking interaction or lignan intake x fiber intake interaction for any of the endpoints (P for interaction 0.44 for all; results not shown).

View this table:
TABLE 3. Rate ratios (RR) for mortality from coronary heart disease (CHD), cardiovascular disease (CVD), cancer, and all causes per 1-SD increase in intake of lignans in 570 men in the Zutphen Elderly Study¹

 
After full adjustment (model 2), MAT was the only individual lignan whose intake was significantly associated with mortality. The intake of MAT was significantly inversely associated with CHD, CVD, and all-cause mortality, and borderline (P = 0.06) with cancer mortality (Table 3). When we did not include the adjustment for ß-carotene and dietary fiber (model 1) we found significant associations between LARI intake and both CHD and all-cause mortality and between SECO intake and CVD mortality.

Tea is the major contributor (47%) to the intake of MAT. Since tea is inversely associated with CHD risk in this population (19, 22), we evaluated whether MAT's protective effect could be attributed to other components in tea, eg, catechins or flavonols. Because of multicollinearity, it was not possible to adjust for these factors in the regression analyses. Therefore, after adjustment for tea consumption, we investigated whether MAT from sources other than tea was associated with mortality. For MAT from sources other than tea, the RRs per 1-SD increase in intake did not differ significantly from those for total MAT intake. In the fully adjusted models RRs (95% CI) were 0.71 (0.52, 0.97) for CHD, 0.79 (0.65, 0.95) for CVD, 0.74 (0.59, 0.94) for cancer, and 0.85 (0.75, 0.96) for all-cause mortality.

We also evaluated the effect of adjustment for the other major dietary sources of MAT (ie, vegetables, bread, coffee, fruit, and wine) by adding these variables one by one to the fully adjusted models. None of these variables substantially changed the RRs (<5% change in RRs), except wine consumption (Table 4). After additional adjustment for wine consumption, the RR (95% CI) for CHD was attenuated from 0.72 (0.53–0.98) to 0.85 (0.61–1.18). The RRs for the other endpoints also were attenuated. Because wine consumption was associated with alcohol consumption and socio-economic status (SES), we evaluated the effect of adjusting for these variables. Additional adjustment for alcohol intake (0, 30, >30 g/d) did not change the associations. Additional adjustment for SES, attenuated the association between the intake of MAT and CHD mortality, but the attenuation was less than that observed after adjustment for wine consumption.

View this table:
TABLE 4. Rate ratios (RR) for mortality from coronary heart disease (CHD), cardiovascular disease (CVD) cancer, and all causes per 1-SD increase in matairesinol intake with additional adjustments for factors associated with matairesinol intake in 570 men in the Zutphen Elderly Study¹

 

DISCUSSION  
In this prospective study of elderly men who were free of CVD and cancer at baseline, total lignan intake was not associated with mortality, after adjustment for potential confounders. However, the intake of MAT was significantly and inversely associated with CHD, CVD, and all-cause mortality and tended toward a significant association with cancer mortality. Additional adjustment for the consumption of wine, an important source of MAT, attenuated these associations, especially the association with CHD mortality.

In previous epidemiologic studies, only the intakes of SECO and MAT were investigated. In the current study, SECO and MAT contributed 16% to the mean total lignan intake. Thus, the inclusion of 4 dietary lignans gives a better representation of the exposure to enterolignans than does SECO or MAT alone. Although arctigenin, 7-hydroxymatairesinol, syringaresinol (3), sesamin (26) and lignins (27) have also been identified as enterolignan precursors, they appear to be less important than the 4 selected precursors, because their degree of conversion is lower, they are not commonly present in foods, or both.

An alternative to the quantification of precursors in the diet is the measurement of enterolignan concentrations in plasma or urine. This approach has the advantage that it takes into account all precursors, as well as their bioavailability, but it also has the disadvantage that those biomarkers reflect mostly short-term intake (28). The dietary history method used in this study reflects habitual dietary intake, which is more relevant to mortality risk.

So far, only a few studies have investigated the relation between lignan intake and CVD. In elderly women in the Netherlands, lignan intake was inversely associated with incident CHD only in former and current smokers (7). This result was not confirmed in the present study. The results with respect to lignan intake contrast with those of a Finnish study of plasma enterolactone concentrations and CVD. In that study, men in the highest quartile of plasma enterolactone had a significantly lower risk of incident myocardial infarction [RR: 0.35; 95% CI: 0.14, 0.88 (29)], and of CHD (0.44; 0.20, 0.96) and total CVD (0.55; 0.29, 1.01) mortality (30). However, in this study some important potential confounders, such as physical activity were not considered.

Especially for men, data on the relation between lignan intake and cancer are still limited. No association was found between lignan intake and testicular cancer (14) or prostate cancer (15). In a case-control study, the intake of SECO plus MAT was inversely associated with lung cancer in men, but not in women. We did not observe an inverse association between total lignan intake and cancer mortality, although men with a high intake of MAT tended to have lower cancer mortality.

We noted consistent associations between the intake of MAT and the various endpoints. This is remarkable, because the intake of MAT is very low compared with that of the other lignans. We assumed that lignans have to be converted to enterolignans to be physiologically active. Under this assumption the prominent role of MAT can be explained only if MAT is more efficiently converted. However, an in vitro fermentation experiment (3) and a subsequent study in rats (31), did not indicate a higher degree of conversion of MAT than of the other 3 major precursors. So, our data suggest that enterolignans may not be the physiologically active compounds.

Support for a direct role of MAT comes from studies that showed that plant lignans were absorbed as such (32). In an in vitro study, MAT inhibited matrix metalloproteinase 7 (MMP-7), whereas SECO showed no inhibition (33). MMP-7 is thought to play a role in tumor invasion and metastasis as well as in atherosclerosis (34). This inhibition of MMP-7 could be a mechanism for a protective effect of MAT against both cancer and CVD, and it may also explain the results from a German case-control study in which breast cancer risk was inversely associated with MAT intake but not with SECO or total lignan intake (35). However, it should be noted that the concentrations used in vitro are much higher than the expected physiologic lignan concentrations.

Because of the observational nature of the present study, residual confounding cannot be fully excluded as an explanation for the observed associations. We have adjusted for age, smoking, physical activity, and the intakes of energy, ß-carotene, and dietary fiber. To gain more insight in potential sources of residual confounding, we have evaluated the effect of adjustment for the major dietary sources of MAT. The associations between MAT intake and mortality were substantially attenuated only by adjustment for wine consumption. The RR for CHD was particularly attenuated. This change could not be attributed to alcohol because adjustment for alcohol intake did not change the RRs. Thus, bioactive compounds other than alcohol found in wine may explain the inverse associations found for MAT.

In summary, we observed no association between total lignan intake and CHD, CVD, cancer, and all-cause mortality in a population of elderly men. Only the intake of MAT, a minor lignan, was inversely associated with all endpoints considered, which suggests that enterolignans are not the active compounds. Associations with MAT were attenuated and no longer statistically significant after additional adjustment for wine consumption. Thus, either MAT or a factor associated with MAT intake, such as wine consumption, may be responsible for the inverse associations of MAT with CVD and cancer.

ACKNOWLEDGMENTS  
PCHH and DK contributed to the study concept and design. IEJM, EJMF, ICWA, and PCHH contributed to the data collection. IEJM performed the statistical data analyses and wrote the manuscript. EJMF and ICWA, HBB, PCHH, and DK contributed to the interpretation of the data and revisions of the manuscript. None of the authors had any financial or personal conflict of interest.

REFERENCES